X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Futil%2FComparison.java;h=2dcbeb5a1d2a076e9b118604510f332b6f3a26ff;hb=e134aeee6802ab52748920ec07220c6a6431b77c;hp=cd98ee74bf560bf426c948950cc1e094809ff3a4;hpb=04c8f7bff663aa469127e9eed4164e02933782f1;p=jalview.git

diff --git a/src/jalview/util/Comparison.java b/src/jalview/util/Comparison.java
index cd98ee7..2dcbeb5 100644
--- a/src/jalview/util/Comparison.java
+++ b/src/jalview/util/Comparison.java
@@ -20,11 +20,11 @@
  */
 package jalview.util;
 
-import jalview.datamodel.SequenceI;
-
 import java.util.ArrayList;
 import java.util.List;
 
+import jalview.datamodel.SequenceI;
+
 /**
  * Assorted methods for analysing or comparing sequences.
  */
@@ -32,6 +32,8 @@ public class Comparison
 {
   private static final int EIGHTY_FIVE = 85;
 
+  private static final int NINETY_NINE = 99;
+
   private static final int TO_UPPER_CASE = 'a' - 'A';
 
   public static final char GAP_SPACE = ' ';
@@ -256,7 +258,7 @@ public class Comparison
    */
   public static final boolean isGap(char c)
   {
-    return (c == GAP_DASH || c == GAP_DOT || c == GAP_SPACE) ? true : false;
+    return c == GAP_DASH || c == GAP_DOT || c == GAP_SPACE;
   }
 
   /**
@@ -275,12 +277,13 @@ public class Comparison
     long ntCount = 0;
     long aaCount = 0;
     long nCount = 0;
+    long ntaCount = 0;
 
     int len = seq.getLength();
     for (int i = 0; i < len; i++)
     {
       char c = seq.getCharAt(i);
-      if (isNucleotide(c) || isX(c))
+      if (isNucleotide(c))
       {
         ntCount++;
       }
@@ -291,6 +294,13 @@ public class Comparison
         {
           nCount++;
         }
+        else
+        {
+          if (isNucleotideAmbiguity(c))
+          {
+            ntaCount++;
+          }
+        }
       }
     }
     /*
@@ -304,6 +314,12 @@ public class Comparison
     }
     else
     {
+      // check for very large proportion of nucleotide and all ambiguity codes
+      if ((ntCount + nCount + ntaCount) * 100 >= NINETY_NINE
+              * (ntCount + aaCount))
+      {
+        return ntCount > 0;
+      }
       return false;
     }
   }
@@ -350,11 +366,13 @@ public class Comparison
    */
   public static boolean isNucleotide(char c)
   {
-    if ('a' <= c && c <= 'z')
-    {
-      c -= TO_UPPER_CASE;
-    }
-    switch (c)
+    return isNucleotide(c, false);
+  }
+
+  public static boolean isNucleotide(char c, boolean countAmbiguity)
+  {
+    char C = Character.toUpperCase(c);
+    switch (C)
     {
     case 'A':
     case 'C':
@@ -363,29 +381,45 @@ public class Comparison
     case 'U':
       return true;
     }
+    if (countAmbiguity)
+    {
+      boolean ambiguity = isNucleotideAmbiguity(C);
+      if (ambiguity)
+        return true;
+    }
     return false;
   }
 
-  public static boolean isN(char c)
+  public static boolean isNucleotideAmbiguity(char c)
   {
-    switch (c)
+    switch (Character.toUpperCase(c))
     {
-    case 'N':
-    case 'n':
+    case 'I':
+    case 'X':
+    case 'R':
+    case 'Y':
+    case 'W':
+    case 'S':
+    case 'M':
+    case 'K':
+    case 'B':
+    case 'H':
+    case 'D':
+    case 'V':
       return true;
+    case 'N': // not counting N as nucleotide
     }
     return false;
   }
 
+  public static boolean isN(char c)
+  {
+    return 'n' == Character.toLowerCase(c);
+  }
+
   public static boolean isX(char c)
   {
-    switch (c)
-    {
-    case 'X':
-    case 'x':
-      return true;
-    }
-    return false;
+    return 'x' == Character.toLowerCase(c);
   }
 
   /**
@@ -456,13 +490,7 @@ public class Comparison
   public static boolean isSameResidue(char c1, char c2,
           boolean caseSensitive)
   {
-    if (caseSensitive)
-    {
-      return (c1 == c2);
-    }
-    else
-    {
-      return Character.toUpperCase(c1) == Character.toUpperCase(c2);
-    }
+    return caseSensitive ? c1 == c2
+            : Character.toUpperCase(c1) == Character.toUpperCase(c2);
   }
 }